Published on: 2026-06-18
When Michael Burry (The Big Short) issued a warning in May regarding a potential AI-driven bubble, many interpreted it as a prediction of an imminent crash. However, his primary concern was the extent to which AI-related trades are now influencing the market’s response to each new data release.
The comparison to the 1999-2000 dot-com era is frequently brought up for a reason. While the internet represented a genuine technological advancement, speculative trading around it became excessive. Burry cautions that current AI market dynamics are beginning to resemble this pattern.
US stocks remain close to record territory. On 8 June, the S&P 500 closed at 7,405.73, less than 3% below its record close from the previous week. Investors are still focused on AI and chip companies, even though tech and semiconductor stocks have been shaky lately.
NVIDIA is different from many dot-com companies in the 90’s, as they offer a real and important product, the chips needed to build an AI system. That gives today’s AI stock rally a stronger base than the internet bubble of the late 1990s, which was often driven by hype. Still, the market is vulnerable because future gains depend on demand for AI chips staying strong. While the rally may continue, the initial phase of rapid gains has likely concluded. Going forward, semiconductor stocks must maintain leadership, and corporate earnings must continue to justify current valuations.
Most bubble warnings focus on how far prices have run. The concern is that investors are paying too much and not taking the risks seriously enough.
Burry’s bigger point is that investors may be looking at every new piece of news mainly through one lens: What it means for AI Stocks.
A strong jobs report helps the rally because it suggests the economy is still healthy, weaker signs, like poor consumer confidence, may be ignored if they do not seem to hurt demand for AI chips or AI spending.
As AI-related stocks keep rising, more buyers jump in, not wanting to miss out. That pushes prices even higher and makes bullish investors feel more confident.
The key question is whether these stocks are still rising because of real new information or simply because they are linked to AI.
The 1999 comparison works because the dot-com bust is easy to read in hindsight. The internet was a real technological shift, yet plenty of internet stocks ran far beyond what their businesses could support.
AI could change the economy in a big way, but stock prices may already be assuming too much good news. Strong demand for chips does not remove that risk.
There are also clear similarities with the late 1990s dot-com boom. Chipmakers and big tech companies are shaping the main market story, while investors who stayed out have watched stocks keep rising without them.
Today’s leading AI and chip companies are not just speculative startups. Many are profitable, well-established businesses with large customer bases and real sales from chips, cloud services, and AI infrastructure.
That makes the current situation very different from 1999, when many dot-com companies had little revenue and were valued mostly on hopes for future growth.
The primary risk lies in current valuations. Investors are paying for anticipated growth that may materialise much later than present prices suggest.
Paul Tudor Jones has also drawn parallels to the late 1990s, noting that current conditions resemble those of 1999, although he acknowledges the rally could persist. Burry emphasises the reality of the risk, while Jones highlights that genuine risks may take years to manifest.
Such warnings can sometimes prove to be accurate but offer little guidance on timing. The timing problem is why overextended markets often continue to rise beyond expectations. Acting prematurely can forfeit potential gains, while inaction may expose investors to sudden reversals.
The wiser move is to wait for clear signs that the rally is weakening, rather than trying to guess the exact moment the market will turn.
Four indicators would suggest that the rally is weakening.
The first is whether markets become more sensitive to economic data and bond yields again. For now, weak data or higher yields may be brushed aside because investors remain focused on AI. But if that same kind of news starts dragging stocks lower, it would suggest investors are becoming less willing to ignore risks outside the AI story.
The second is whether semiconductor stocks begin to lose leadership. Chips are at the heart of the AI trade, which is why Burry focused on the Philadelphia Semiconductor Index. If chip stocks start falling behind while the Nasdaq or S&P 500 continues to rise, it may mean the rally is losing some of its support.
The third is market breadth, which measures how many stocks are taking part in the rally. A healthy rally usually includes a wide range of companies. If gains become concentrated in only a few AI-related names, the market may be more fragile than it appears.
On 8 June, only about 53.5% of S&P 500 stocks were above their 50-day moving average. MarketWatch also cited FactSet data showing more than 300 index members falling even as the S&P 500 traded higher. It is not collapse-level breadth, but it shows how much the headline index can depend on a smaller group of winners.
When only a few AI-related stocks are driving the index, the market can look stronger than it really is. If those leaders start to weaken, the rest of the market may not offer enough support.
The fourth test is how the market handles earnings and capital spending.
In a robust rally, strong financial results are typically rewarded. In an overextended market, such results may be dismissed as already reflected in prices. This principle also applies to major cloud companies’ capital expenditures; any reduction or hesitation in data centre and chip spending could prompt investors to reassess the growth expectations embedded in AI-related shares.
Burry has not specified a timeline, nor can anyone. However, careful observation of market behaviour can provide valuable insights for investors.
AI remains a primary driver of investor activity, and markets heavily concentrated in a single theme can persist for extended periods. However, the margin for error has diminished. Disappointments that were previously overlooked may now have a more pronounced negative impact, with the indicators likely to reveal early signs of weakness.
Selling or changing your portfolio just because of warnings may hurt you over time. The main lesson is not that the market must fall, but that investors should not assume AI-related stocks will keep rising without risk.
